Summary:
Anthropic is working with the U.S. government to expand access to its most advanced AI models, allowing a select group of agencies to test the startup's cutting-edge cyber capabilities. According to a statement released on Tuesday, the San Francisco-based company will allow proven agencies to use its most capable models, including Claude Opus 5.5, Claude Sonnet 5.5, Claude Mythos 5.1, as well as new models coming in the future. The statement said authorized agencies can conduct "high-risk offensive testing" of security systems designed to protect critical infrastructure such as power grids, banks and flight operations systems.
Anthropic passed the "Glasswing Project" (Project Glasswing) in April this year to limit the opening of the Mythos model to US government agencies, financial institutions and large software vendors. The release of Mythos was a watershed moment in the field of cybersecurity because the model was able to automate parts of the work of advanced hackers. Since then, a series of incidents involving Anthropic and OpenAI models accidentally intruding into government, corporate and nonprofit systems have raised concerns about the security of the technology.
Anthropic said on Tuesday that the latest round of limited openings will cover all members of the "Glass Wing Project" and that the company will work with the U.S. government to review all new institutions.
Comments